PS 53-197 - Enhanced minirhizotron image contrast from multispectral image analyses

Friday, August 12, 2016
ESA Exhibit Hall, Ft Lauderdale Convention Center
Brian Scott, School of Life Sciences, Arizona State University, Tempe, AZ, Josh Haussler, School of Life Sciences, Arizona State University and Osvaldo E. Sala, School of Life Sciences and School of Sustainability, Arizona State University, Tempe, AZ
Background/Question/Methods

Minirhizotrons are widely used as a non-invasive method to monitor and analyze plant roots. Current methods for analyzing MR images are labor intensive, requiring a significant time investment to manually identify roots. Methods to improve processing time and accuracy often focus on image analysis software. We used multi-spectral imaging techniques to improve the image contrast prior to image analysis. Root samples were analyzed using a Leica Microsystems TCS SP5 confocal laser microscope. Fluorescent spectral scans were generated with 4 incident light wavelengths: 405nm, 488nm, 561nm and 633nm. Of these, the 488nm laser produced the best contrast and was carried forward to field trials. To approximate the 488nm incident laser light source, 470nm peak wavelength LED bulbs were substituted for the full spectrum incandescent bulbs. A 550 nm high performance longpass optical filter was used to remove the incident light source (< 0.5% passing) while allowing the fluorescent light from the active (live) roots to reach the camera lens (> 50% passing). Laboratory and field tests were conducted with a modified Bartz MR and Image J was used for image processing. Roots and soil were collected from a research site in southern New Mexico.

Results/Conclusions

The purpose of this work was to determine if MR root-soil image contrast could be improved. In our trials, the use of a narrow spectrum incident light sources provided a strong image contrast between live roots and background soil. Images did not require manual identification, which accelerates image processing time. The multispectral analysis was performed on live and dead roots at four incident wavelengths. In our root-soil combination, each wavelength provided a qualitatively different response. The 561nm source illuminated both the live and dead root stalks and resulted in a relatively strong fluorescent signal. Live root images also showed the root hairs (dead roots do not have root hairs). The 633nm source resulted in a weak signal. Root hairs were less visible and the live root stalk had a significantly stronger signal than the dead root stalk. The 405nm source provided the strongest signal, illuminating dead and live root stalks and hairs. The 488nm source provided a moderate signal and illuminated primarily root hairs. Preferentially illuminating root hairs is advantageous because they are more difficult to identify manually and are more closely associated with plant activity. Although results are preliminary, this works points to a potentially simple, low cost modification to existing MR equipment.